Understanding Inertia and Trait-State Influences in Longitudinal Psychological Processes

Google map

Abstract: Inertia in multivariate longitudinal processes captures the carryover of states over time, yet its interpretation and implications often remain unclear, particularly when dealing with irregularly spaced data. This talk explores two methodological innovations that advance the understanding of inertia and trait-state influences in longitudinal psychological research. Study 1 introduces new methods for uncertainty quantification in continuous-time mediation models, addressing a critical gap in the field. While continuous-time vector autoregressive (CT-VAR) models overcome the time-interval dependency inherent in discrete-time approaches, they have lacked tools to estimate the uncertainty of direct, indirect, and total effects. We present novel approaches using the delta and Monte Carlo methods, demonstrating their implementation in the R package cTMed. Study 2 introduces the Common and Unique Latent Transition Analysis (CULTA), a new model that integrates latent transition analysis with trait-state modeling to distinguish enduring trait patterns from transient state fluctuations. Applied to transdermal alcohol concentration data from heavy episodic drinkers, CULTA uncovers nuanced intoxication profiles and highlights risk and protective factors, providing actionable insights for early screening and prevention strategies. Other unresolved challenges and future directions are summarized.